We used the CLUMPY torus models and a Bayesian approach to fit the infrared spectral energy distributions and ground-based high angular resolution mid-infrared spectroscopy of 13 nearby Seyfert galaxies. This allowed us to put tight constraints on torus model parameters such as the viewing angle i, the radial thickness of the torus Y, the angular size of the cloud distribution σtorus, and the average number of clouds along radial equatorial rays N 0. We found that the viewing angle i is not the only parameter controlling the classification of a galaxy into type 1 or type 2. In principle, type 2s could be viewed at any viewing angle i as long as there is one cloud along the line of sight. A more relevant quantity for clumpy media is the probability for an active galactic nucleus (AGN) photon to escape unabsorbed. In our sample, type 1s have relatively high escape probabilities, P esc ~ 12%-44%, while type 2s, as expected, tend to have very low escape probabilities. Our fits also confirmed that the tori of Seyfert galaxies are compact with torus model radii in the range 1-6 pc. The scaling of the models to the data also provided the AGN bolometric luminosities L bol(AGN), which were found to be in good agreement with estimates from the literature. When we combined our sample of Seyfert galaxies with a sample of PG quasars from the literature to span a range of L bol(AGN) ~ 1043-1047 erg s-1, we found plausible evidence of the receding torus. That is, there is a tendency for the torus geometrical covering factor to be lower (f 2 ~ 0.1-0.3) at high AGN luminosities than at low AGN luminosities (f 2 ~ 0.9-1 at ~1043-1044 erg s-1). This is because at low AGN luminosities the tori appear to have wider angular sizes (larger σtorus) and more clouds along radial equatorial rays. We cannot, however, rule out the possibility that this is due to contamination by extended dust structures not associated with the dusty torus at low AGN luminosities, since most of these in our sample are hosted in highly inclined galaxies.